Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Deploy Figma-powered applications to Vercel, Cloud Run, and Fly.io. Use when deploying webhook receivers, design token APIs, or Figma-connected web apps to production platforms. Trigger with phrases like "deploy figma", "figma Vercel", "figma production deploy", "figma Cloud Run".
.claude/skills/jeremylongshore-figma-deploy-integration/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 48% | 0% |
Deploy Figma webhook receivers and design API services to production platforms with proper secret management and health checks.
bash# Store Figma secrets vercel env add FIGMA_PAT production vercel env add FIGMA_WEBHOOK_PASSCODE production # Deploy vercel --prod
typescript// api/webhooks/figma.ts (Vercel serverless function) import { NextRequest, NextResponse } from 'next/server'; import crypto from 'crypto'; export async function POST(req: NextRequest) { const payload = await req.json(); // Verify passcode const expected = process.env.FIGMA_WEBHOOK_PASSCODE!; const received = payload.passcode || ''; const a = Buffer.from(received); const b = Buffer.from(expected); // timingSafeEqual throws on length mismatch — guard first if (a.length !== b.length || !crypto.timingSafeEqual(a, b)) { return NextResponse.json({ error: 'Invalid passcode' }, { status: 401 }); } // Process webhook event switch (payload.event_type) { case 'FILE_UPDATE': console.log(`File updated: ${payload.file_name} (${payload.file_key})`); // Trigger token re-sync, invalidate cache, etc. break; case 'FILE_COMMENT': console.log(`New comment on ${payload.file_name}`); break; case 'LIBRARY_PUBLISH': console.log(`Library published: ${payload.file_name}`); break; } return NextResponse.json({ received: true }); } export const config = { maxDuration: 10 };
dockerfileFROM node:20-slim WORKDIR /app COPY package*.json ./ RUN npm ci --omit=dev COPY dist/ ./dist/ ENV PORT=8080 CMD ["node", "dist/server.js"]
bashPROJECT_ID="${GOOGLE_CLOUD_PROJECT}" SERVICE="figma-token-api" REGION="us-central1" # Store PAT in Secret Manager echo -n "${FIGMA_PAT}" | gcloud secrets create figma-pat --data-file=- # Build and deploy gcloud builds submit --tag gcr.io/$PROJECT_ID/$SERVICE gcloud run deploy $SERVICE \ --image gcr.io/$PROJECT_ID/$SERVICE \ --region $REGION \ --platform managed \ --set-secrets="FIGMA_PAT=figma-pat:latest" \ --allow-unauthenticated \ --max-instances=5 \ --timeout=30s
toml# fly.toml app = "figma-webhook-service" primary_region = "iad" [env] NODE_ENV = "production" [http_service] internal_port = 3000 force_https = true auto_stop_machines = "suspend" auto_start_machines = true min_machines_running = 1 [[http_service.checks]] grace_period = "10s" interval = "30s" method = "GET" path = "/health" timeout = "5s"
bashfly secrets set FIGMA_PAT=figd_your-token fly secrets set FIGMA_WEBHOOK_PASSCODE=your-passcode fly deploy
typescript// src/health.ts -- works on any platform import { figmaFetch } from './figma-client'; export async function healthHandler(req: Request): Promise<Response> { const start = Date.now(); try { const res = await fetch('https://api.figma.com/v1/me', { headers: { 'X-Figma-Token': process.env.FIGMA_PAT! }, signal: AbortSignal.timeout(5000), }); return Response.json({ status: res.ok ? 'healthy' : 'degraded', figma: { authenticated: res.ok, latencyMs: Date.now() - start, }, timestamp: new Date().toISOString(), }); } catch { return Response.json({ status: 'unhealthy', figma: { authenticated: false, latencyMs: Date.now() - start }, }, { status: 503 }); } }
| Issue | Cause | Solution | |-------|-------|----------| | Secret not found in runtime | Wrong env name | Verify with platform CLI (vercel env ls) | | Webhook timeout | Processing too slow | Return 200 immediately, process async | | Cold start latency | Serverless cold boot | Use Fly.io min_machines_running: 1 or Cloud Run min instances | | Health check fails | PAT expired | Rotate token via platform secret management |
Deploy the webhook receiver to Vercel (Step 1) and verify end-to-end:
bashvercel deploy --prod curl -s -X POST https://figma-hooks.example.vercel.app/api/figma/webhook \ -H 'Content-Type: application/json' \ -d '{"event_type":"PING","passcode":"'"${WEBHOOK_PASSCODE}"'"}' # 200 {"ok":true}
Then point Figma at it and watch a real event arrive:
textPOST /api/figma/webhook 200 event=FILE_UPDATE file=AbC123 triggered_by=mia.designer
Probe the deployed health endpoint (Step 4) — it checks Figma reachability, not just process-up:
bashcurl -s https://figma-hooks.example.vercel.app/api/health | jq . # {"status":"ok","figma_api":"reachable","uptime_s":86400}
Cloud Run and Fly.io equivalents: references/google-cloud-run-design-token-api.md, references/fly-io-persistent-webhook-service.md.
For webhook handling, see figma-webhooks-events.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,038 | 10,660 | -29% | 1 | 1 | 0% | 3,061 | 3,799 | +24% | 0 | 0 | — |
case-02 | fail→fail | 14,839 | 14,425 | -3% | 1 | 1 | 0% | 2,892 | 3,912 | +35% | 0 | 0 | — |
case-03 | fail→fail | 17,235 | 14,086 | -18% | 1 | 1 | 0% | 3,719 | 4,001 | +8% | 0 | 0 | — |
case-04 | fail→pass | 6,488 | 2,805 | -57% | 1 | 1 | 0% | 1,266 | 2,162 | +71% | 0 | 0 | — |
case-05 | pass→pass | 13,614 | 8,516 | -37% | 1 | 1 | 0% | 2,341 | 3,150 | +35% | 0 | 0 | — |
case-06 | pass→pass | 14,002 | 3,215 | -77% | 1 | 1 | 0% | 2,038 | 2,168 | +6% | 0 | 0 | — |
case-07 | pass→pass | 15,821 | 3,171 | -80% | 1 | 1 | 0% | 2,330 | 2,081 | -11% | 0 | 0 | — |
case-08 | pass→pass | 13,133 | 5,399 | -59% | 1 | 1 | 0% | 2,251 | 2,567 | +14% | 0 | 0 | — |
case-09 | pass→pass | 6,449 | 3,518 | -45% | 1 | 1 | 0% | 1,222 | 2,336 | +91% | 0 | 0 | — |
case-10 | fail→pass | 11,008 | 2,901 | -74% | 1 | 1 | 0% | 1,917 | 2,235 | +17% | 0 | 0 | — |
case-11 | pass→pass | 16,569 | 11,264 | -32% | 1 | 1 | 0% | 2,677 | 3,675 | +37% | 0 | 0 | — |
case-12 | fail→pass | 13,410 | 3,780 | -72% | 1 | 1 | 0% | 2,050 | 2,374 | +16% | 0 | 0 | — |
case-13 | pass→pass | 8,822 | 3,077 | -65% | 1 | 1 | 0% | 1,348 | 2,217 | +64% | 0 | 0 | — |
case-14 | pass→pass | 12,697 | 3,107 | -76% | 1 | 1 | 0% | 1,795 | 2,275 | +27% | 0 | 0 | — |
case-15 | fail→pass | 13,356 | 6,250 | -53% | 1 | 1 | 0% | 2,098 | 3,095 | +48% | 0 | 0 | — |
case-16 | pass→pass | 13,348 | 12,968 | -3% | 1 | 1 | 0% | 2,188 | 3,934 | +80% | 0 | 0 | — |
case-17 | pass→pass | 16,443 | 12,663 | -23% | 1 | 1 | 0% | 2,481 | 4,174 | +68% | 0 | 0 | — |
case-18 | pass→pass | 3,198 | 2,737 | -14% | 1 | 1 | 0% | 580 | 2,072 | +257% | 0 | 0 | — |
case-19 | pass→pass | 4,315 | 2,488 | -42% | 1 | 1 | 0% | 564 | 1,993 | +253% | 0 | 0 | — |
case-20 | pass→pass | 734,259 | 13,785 | -98% | 1 | 1 | 0% | 2,447 | 4,008 | +64% | 0 | 0 | — |
case-21 | pass→pass | 18,742 | 12,289 | -34% | 1 | 1 | 0% | 3,971 | 4,321 | +9% | 0 | 0 | — |
case-22 | pass→pass | 12,357 | 12,096 | -2% | 1 | 1 | 0% | 2,452 | 4,122 | +68% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +23 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.